Knowledge tracing is one of the key research areas for empowering\npersonalized education. It is a task to model students' mastery level of a\nknowledge component (KC) based on their historical learning trajectories. In\nrecent years, a recurrent neural network model called deep knowledge tracing\n(DKT) has been proposed to handle the knowledge tracing task and literature has\nshown that DKT generally outperforms traditional methods. However, through our\nextensive experimentation, we have noticed two major problems in the DKT model.\nThe first problem is that the model fails to reconstruct the observed input. As\na result, even when a student performs well on a KC, the prediction of that\nKC's mastery level decreases instead, and vice versa. Second, the predicted\nperformance for KCs across time-steps is not consistent. This is undesirable\nand unreasonable because student's performance is expected to transit gradually\nover time. To address these problems, we introduce regularization terms that\ncorrespond to reconstruction and waviness to the loss function of the original\nDKT model to enhance the consistency in prediction. Experiments show that the\nregularized loss function effectively alleviates the two problems without\ndegrading the original task of DKT.\n